Efficient compression and transport of video over a network
The disclosed invention is a enhancement to the Video Encoder component of the MPEG standard to improve both the efficiency and quality of the video presentation at the display device. The inventions described below pertain to the enhancement of video compression technology. In general, they are compatible with MPEG video standards (as in current art), but represent novel enhancements thereof. The above inventions apply to several transmission media and networks including cable, DSL, wireless, satellite and IP networks. Another category of invention in this disclosure enhances networking performance by recognizing that the transmitted information is video, rather than data. This invention can be scaled to other classes of multimedia data such as speech or audio or images.
1. A method for allocating bits among portions of a frame in a video encoding implementation, said allocation intended to attain a constant bit rate per frame, said method comprising:
in a processor coupled to a memory, determining a particular class of visual sensitivity for a portion of the frame among a plurality of visual sensitivity classes;
in the processor, determining a number of bits associated with the particular class; and
in the processor, establishing a number of bits to be transmitted for the portion based on the number of bits associated with the particular class, wherein a sum over the plurality of classes of a product of the number of bits associated with each class and the probability that the portion belongs to the class is substantively equal to the constant bit rate.
2. A method for adaptive bit allocation, comprising:
in a processor coupled to a memory, repeatedly and dynamically determining a plurality of parameters associated with a frame of a video stream, the plurality of parameters comprising at least a particular class of visual sensitivity for a portion of the frame among a plurality of visual sensitivity classes; and
in the processor, establishing a number of bits to be transmitted for the frame of the video stream based on the determined plurality of parameters, including at least the particular class of visual sensitivity for a portion of the frame;
wherein determining a plurality of parameters associated with a frame comprises:
determining a first number of bits for a frame to achieve a target bit rate according to a first procedure that produces values for a number of header bits, a number of coded zeros and a number of DCT bits; and
determining a second number of bits for the frame, wherein the determination is based on an actual number of coded zeros for the frame to match the number of DCT bits.
3. The method of claim 2 , wherein allocating a bit to the frame based on the determined parameter comprises allocating the second number of bits to the frame.
4. A method for adaptive bit allocation, comprising:
in a processor coupled to a memory, repeatedly and dynamically determining a plurality of parameters associated with a frame of a video stream, the plurality of parameters comprising at least a particular class of visual sensitivity for a portion of the frame among a plurality of visual sensitivity classes; and
in the processor, establishing a number of bits to be transmitted for the frame of the video stream based on the determined plurality of parameters, including at least the particular class of visual sensitivity for a portion of the frame;
wherein determining a plurality of parameters associated with a frame comprises:
determining a relationship that relates a variance in a frame to a number of bits for the frame for each type of frame among an I-frame type, P-frame-type and B-frame type in the group of pictures;
determining a variance corresponding to each type of frame in the group of pictures; and
determining a number of bits for each type of frame based at least in part on the relationship and the variance of each type of frame and the number of frames of each type of frame.
5. The method of claim 4 , wherein allocating a bit to the frame based on the determined parameter comprises allocating the number of bits determined for each type of frame.
6. A method for adaptive bit allocation, comprising:
in a processor coupled to a memory, repeatedly and dynamically determining a plurality of parameters associated with a frame of a video stream, the plurality of parameters comprising at least a particular class of visual sensitivity for a portion of the frame among a plurality of visual sensitivity classes,
wherein determining a plurality of parameters associated with a frame comprises determining whether a macroblock within the frame is associated with motion faster than a threshold speed for tracking by a human observer; and
in the processor, establishing a number of bits to be transmitted for the frame of the video stream based on the determined plurality of parameters, including at least the particular class of visual sensitivity for a portion of the frame,
wherein, establishing a number of bits to be transmitted for the frame based on the determined parameter comprises using fewer bits to represent Discrete Cosine Transform amplitudes for blocks in the macroblock, wherein fewer bits are used if the determination indicates that the macroblock is associated motion faster than the threshold speed.
7. A method for adaptive bit allocation, comprising:
in a processor coupled to a memory, repeatedly and dynamically determining a plurality of parameters associated with a frame of a video stream, the plurality of parameters comprising at least a particular class of visual sensitivity for a portion of the frame among a plurality of visual sensitivity classes; and
in the processor, allocating a bit of the memory to the frame of the video stream based on the determined plurality of parameters, including at least the particular class of visual sensitivity for a portion of the frame;
wherein determining a plurality of parameters associated with a frame comprises:
determining a particular class of visual sensitivity for a portion of the frame among a plurality of visual sensitivity classes, and
determining a number of bits associated with the particular class; and
wherein allocating a bit to the frame based on the determined parameter comprises allocating a number of bits for the portion based on the number of bits associated with the particular class, wherein a sum over the plurality of classes of a product of the number of bits associated with each class and the probability that the portion belongs to the class is substantively equal to the constant bit rate.
8. A system for adaptive bit allocation, comprising:
a memory; and
a processor coupled to the memory and being operable to determine a parameter associated with a frame and allocating a bit to the frame based on the determined parameter, the parameter comprising at least a particular class of visual sensitivity for a portion of the frame among a plurality of visual sensitivity classes;
wherein determining a parameter associated with a frame comprises:
determining a first number of bits for a frame to achieve a target bit rate according to a first procedure that produces values for a number of header bits, a number of coded zeros and a number of DCT bits; and
determining a second number of bits for the frame, wherein the determination is based on an actual number of coded zeros for the frame to match the number of DCT bits.
9. The system of claim 8 , wherein allocating a bit to the frame based on the determined parameter comprises allocating the second number of bits to the frame.
10. A system for adaptive bit allocation, comprising:
a memory; and
a processor coupled to the memory and being operable to determine a parameter associated with a frame and allocating a bit to the frame based on the determined parameter, the parameter comprising at least a particular class of visual sensitivity for a portion of the frame among a plurality of visual sensitivity classes;
wherein determining a parameter associated with a frame comprises:
determining a relationship that relates a variance in a frame to a number of bits for the frame for each type of frame among an I-frame type, P-frame-type and B-frame type in the group of pictures;
determining a variance corresponding to each type of frame in the group of pictures; and
determining a number of bits for each type of frame based at least in part on the relationship and the variance of each type of frame and the number of frames of each type of frame.
11. The system of claim 10 , wherein allocating a bit to the frame based on the determined parameter comprises allocating the number of bits determined for each type of frame.
12. A system for adaptive bit allocation, comprising:
a memory; and
a processor coupled to the memory and being operable to determine a parameter associated with a frame and allocating a bit to the frame based on the determined parameter, the parameter comprising at least a particular class of visual sensitivity for a portion of the frame among a plurality of visual sensitivity classes;
wherein determining a parameter associated with a frame comprises determining whether a macroblock within the frame is associated with motion faster than a threshold speed for tracking by a human observer; and
wherein allocating a bit to the frame based on the determined parameter comprises using fewer bits to represent Discrete Cosine Transform amplitudes for blocks in the macroblock, wherein fewer bits are used if the determination indicates that the macroblock is associated motion faster than the threshold speed.
13. A system for adaptive bit allocation, comprising:
a memory; and
a processor coupled to the memory and being operable to determine a parameter associated with a frame and allocating a bit to the frame based on the determined parameter, the parameter comprising at least a particular class of visual sensitivity for a portion of the frame among a plurality of visual sensitivity classes;
wherein determining a parameter associated with a frame comprises:
determining a particular class of visual sensitivity for a portion of the frame among a plurality of visual sensitivity classes; and
determining a number of bits associated with the particular class.
14. The system of claim 13 , wherein allocating a bit to the frame based on the determined parameter comprises allocating a number of bits for the portion based on the number of bits associated with the particular class, wherein a sum over the plurality of classes of a product of the number of bits associated with each class and the probability that the portion belongs to the class is substantively equal to the constant bit rate.